Design and Application of a Screening Set for Monophosphine Lig-ands in Metal Catalysis

Abstract

In reaction discovery, the search space of discrete reaction parameters such as catalyst structure is often not explored systematically. We have developed a tool set to aid the search of optimal catalysts in the context of phosphine ligands. A virtual library, kraken, that is representative of the monodentate P(III)-ligand chemical space was utilized as the basis to represent the discrete ligands as continuous variables. Using dimensionality reduction and clustering techniques, we suggested a Phosphine Optimization Screening Set (PHOSS) of 32 commercially available ligands that samples this chemical space completely and evenly. We present the application of this screening set in the identification of active catalyst for various cross-coupling reactions and how well-distributed sampling of the chemical space facilitates identification of active catalysts. Furthermore, we demonstrate how proximity in ligand space can be a useful guide to further explore ligands when very few active catalysts are known.

Advancing Base Metal Catalysis through Data Science: Insight and Predictive Models for Ni-Catalyzed Borylation through Supervised Machine Learning

“…These data suggest that additional training data sets should be evaluated with P­( p -CF 3 -Ph) 3 in the future for augmentation of both the ML models and the identification of superior reactivity thresholds. This clearly reflects common challenges facing the use of data science and machine learning workflows to design experiments that incorporate ideal diversity across chemical reaction space for class variables . Nevertheless, we have shown that an iterative process can allow appropriate course correction to improve models throughout their lifecycle.…”

“…In addition to predicting reaction outcomes for substrates not seen by the model, it would be advantageous to be able to predict reactivity for ligands that the model was not trained on and identify ideal ligand candidates for Ni-catalyzed borylation. A proximity-based ligand space guided search to select 12 new ligands (Figure , red circles) for studying the ML models trained on the original 24 ligands (Figure , blue circles) was performed . The selected ligand set includes eight ligands expected to perform well based on Kraken values (CM-Phos, CPhos, CX-POMeCy, Cy-JohnPhos-OMe, PPh 2 ( o -Anis), PPhCy 2 , RuPhos, and S-Phos) and four ligands expected to perform poorly (PEt 3 , PBn 3 , CX-PInCy, and P­( p -CF 3 -Ph) 3 ).…”

A Comprehensive Discovery Platform for Organophosphorus Ligands for Catalysis

“…The kraken tool may enable informed catalyst design based on organophosphorus ligands, facilitate the optimization of reaction process parameters, inspire new ligand choices, and promote the synthesis of new organophosphorus compounds. The database and tools reported herein are currently being applied to enhance reaction optimization − and mechanistic workflows . The open-source nature of our codes, as well as the open database, is designed to be extended by others, and we welcome further contributions by the community.…”

Univariate classification of phosphine ligation state and reactivity in cross-coupling catalysis

“…We also collected data on Ni-catalyzed Csp 2 –Csp 2 Suzuki-Miyaura coupling (SMC) reactions with aryl chlorides using high-throughput experimentation. The ligand set was assembled from a combined Doyle and Merck Sharp & Dohme inventory to encompass 90 monodentate phosphines that are included in the kraken virtual library ( 21 ), with k -means clustering applied to enable a selection of structures that broadly covered phosphine chemical space ( 27 ). These ligands were evaluated for four cross-coupling reactions wherein the aryl halide and aryl boronic acid were altered (reactions II to V; Fig.…”